Changed horizontal mirror logic in k44. Fixes #851

[rp-jt]

Compared the mirror logic we had for k44 (ghf ^ pre_hf ^ hmir_eff ) with the one in Mame.

Both tables of truth were slightly different, so I added a small change. The pre_hflip is influenced by the horizontal mirroring, but not the other way, and that was causing the problem in #851. This commit fixes #851

Maybe a similar approach with the vertical flip might help with #724, but I need to check it more thoroughly. I tried changing vflip definition in the same way and the scenes stayed the same

[jotego]

It is good. Note that the suffix _eff in signals stands for effective, so it is meant to be used instead of the original signal. In this case, it ends up used together with the original. No big deal, though.